03.12.2012 Views

Semantic Web-Based Information Systems: State-of-the-Art ...

Semantic Web-Based Information Systems: State-of-the-Art ...

Semantic Web-Based Information Systems: State-of-the-Art ...

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

Corcho & Gómez-Pérez<br />

Figure 11. Task decomposition <strong>of</strong> activity 4 (implementation <strong>of</strong> <strong>the</strong> ontology translation<br />

system)<br />

lar to <strong>the</strong> lex (Lesk, 1975) and yacc (Johnson, 1975) languages used for compiler<br />

construction, and that ODESem is based on common rule-based systems.<br />

As in <strong>the</strong> design activity, <strong>the</strong> tasks inside this implementation activity are divided<br />

in groups — four, in this case — as shown in Figure 11.<br />

The goal <strong>of</strong> <strong>the</strong> first task is to select reusable pieces <strong>of</strong> code from <strong>the</strong> declarative<br />

specifications <strong>of</strong> o<strong>the</strong>r ontology translation systems. These pieces <strong>of</strong> code are selected<br />

on <strong>the</strong> basis <strong>of</strong> <strong>the</strong> results obtained from <strong>the</strong> first task <strong>of</strong> <strong>the</strong> design activity<br />

and can be related to any <strong>of</strong> <strong>the</strong> four translation layers.<br />

The next five tasks are grouped toge<strong>the</strong>r and should be performed almost in parallel,<br />

as shown in <strong>the</strong> figure. In <strong>the</strong>se tasks, s<strong>of</strong>tware and knowledge engineers actually<br />

must implement <strong>the</strong> transformations at <strong>the</strong> four layers — lexical, syntax, semantic,<br />

and pragmatic — and <strong>the</strong> additional transformations described in task 3.6. Unlike<br />

in <strong>the</strong> design activity, we propose to start with <strong>the</strong> low-level transformations (those<br />

at <strong>the</strong> lexical and syntax layers) and continue with <strong>the</strong> more abstract (and difficult)<br />

ones. The reason for <strong>the</strong> task ordering suggested is that <strong>the</strong> semantic and pragmatic<br />

transformation implementations usually need to take into account <strong>the</strong> specific implementations<br />

at <strong>the</strong> lexical and syntax layers. We are currently developing automatic<br />

tools that transform <strong>the</strong> declarative specifications in ODELex, ODESyntax, and<br />

ODESem into Java code.<br />

In task 4.7 — declarative specification processing and integration — <strong>the</strong> s<strong>of</strong>tware<br />

engineer is in charge <strong>of</strong> transforming <strong>the</strong> previous declarative implementations<br />

at all levels, plus <strong>the</strong> additional transformations, into actual running code, which<br />

will perform <strong>the</strong> translations as specified in <strong>the</strong> previous code. In addition, <strong>the</strong><br />

s<strong>of</strong>tware engineer has to integrate <strong>the</strong> resulting ontology translation system into<br />

ano<strong>the</strong>r information system (e.g., an ontology tool), if required. Given that most<br />

Copyright © 2007, Idea Group Inc. Copying or distributing in print or electronic forms without written permission <strong>of</strong><br />

Idea Group Inc. is prohibited.

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!